ML-lab

This repository contains the programs that I worked out in Machine Learning Laboratory.

Index

• Lab 1: Introduction to EDA
  • Introduction to Matplotlib and Seaborn packages
  • Introduction to Numpy and Pandas package
  • Insurance Data Analysis
  • Iris Data Analysis
• Lab 2: Principal Component Analysis
  • About PCA
  • PCA based dimensionality reduction on Wine Dataset
  • PCA using algorithm steps without sklearn
  • PCA using sklearn on Iris Dataset
  • PCA - Wine Quality Classification
• Lab 3: K-Nearest Neighbors
  • About KNN
  • KNN using Iris Dataset
• Lab 4: Linear Discriminant Analysis and Linear Regression
  • About LDA and LR
  • LDA on Iris Dataset
  • LR on Single Dataset (Iris)
  • LR on Multiple Datasets (Iris, Wine)
• Lab 5: Logistic Regression
  • About LR
  • LR on Multiple Datasets (Iris, Wine)
• Lab 6: Naive Bayes Classifier
  • About NBC
  • NBC - Basic Program
  • NBC on Iris Dataset
  • NBC Sentiment Analysis on IMDB Dataset (Short program) (for exam)
  • NBC Sentiment Analysis on IMDB Dataset (Detailed program) (for understanding)

---

• Lab 7, 8: Support Vector Machine
  • About SVM
  • SVM Program on Breast Cancer Dataset
  • SVM Program for Linear and non-Linear Classification Tasks on Breast Cancer Dataset
  • SVM Program for Linear and non-Linear Classification Tasks on Iris Dataset
• Lab 9: Multi-Layer Feed Forward Neural Network and Regularization Techniques
  • About MLFFNN and Regularization Techniques
  • MLFFNN on Breast Cancer Dataset (Text) (not for exam)
  • MLFFNN on MNIST Dataset (Image) - Short
  • MLFFNN on MNIST Dataset (Image) - Full
  • Regularization Techniques on Breast Cancer Dataset (Text) (not for exam)
  • Regularization Techniques on MNIST Dataset (Image)
  • Regularization Techniques with comparision on Diabetes Dataset (Text) (Alternative program)
  • Regularization Techniques on Obesity Classification Dataset (Text)
• Lab 10, 11: Artificial Neural Network, Convolutional Neural Network; Hidden Markov Model based techinques (Viterbi Algorithm, Trellis, Long Short Term Memory)
  • About ANN, CNN & HMM
  • ANN on CIFAR10 Dataset
  • CNN on CIFAR10 Dataset (Alternative program)
  • HMM - Viterbi Algorithm on own Weather Dataset (Alternative program)
  • HMM - Viterbi Algorithm (with Trellis) on Iris Dataset
  • HMM - LSTM Algorithm on Iris Dataset (not for exam)
• Complete ML Codes Program

Prerequisites

Python and packages in requirements.txt file installed.

Note

You can install all the required packages using the command pip install -r requirements.txt.

Working with Conda

If you are using conda to manage your environments, you can create a new environment for this repository with the command conda create -n ml-lab and activate it with the command conda activate ml-lab.

Tip

For faster environment solving in Conda, I would suggesting using the libmamba solver. You can set it as the default solver using the command conda config --set solver libmamba.

Then, you can install all the required packages using the command conda install --file requirements.txt.

Container Image

Alternatively, you can use the container image I created with all the packages preinstalled.

You can install it in Distrobox with the command distrobox create -i ghcr.io/kbdharun/ml-lab-image:latest -n ml and use it with the command distrobox enter ml.

Additionally, you can verify the authenticity of the container image using cosign (download the cosign.pub file from here and execute the following command):

cosign verify --key cosign.pub ghcr.io/kbdharun/ml-lab-image:latest